LED light sources (i.e., LED lamps or, more familiarly, LED “light bulbs”) provide an energy-efficient alternative to traditional types of light sources, but typically require specialized circuitry to properly power the LED(s) within the light source. As used herein, the terms LED light sources, lamps, and/or bulbs refer to systems that include LED driver and support circuitry (the “LED module”) as well as the actual LED(s). For LED light sources to gain wide acceptance in place of traditional light sources, their support circuitry must be compatible with as many types of existing lighting systems as possible. For example, incandescent bulbs may be connected directly to an AC mains voltage, halogen-light systems may use magnetic or electronic transformers to provide 12 or 24 VAC to a halogen bulb, and other light sources may be powered by a DC current or voltage. Furthermore, AC mains voltages may vary country-by-country (60 Hz in the United States, for example, and 50 Hz in Europe).
Current LED light sources are compatible with only a subset of the above types of lighting system configurations and, even when they are compatible, they may not provide a user experience similar to that of a traditional bulb. For example, an LED replacement bulb may not respond to a dimmer control in a manner similar to the response of a traditional bulb. One of the difficulties in designing, in particular, halogen-replacement LED light sources is compatibility with the two kinds of transformers (i.e., magnetic and electronic) that may have been originally used to power a halogen bulb. A magnetic transformer consists of a pair of coupled inductors that step an input voltage up or down based on the number of windings of each inductor, while an electronic transformer is a complex electrical circuit that produces a high-frequency (i.e., 100 kHz or greater) AC voltage that approximates the low-frequency (60 Hz) output of a magnetic transformer. FIG. 1 is a graph 100 of an output 102 of an electronic transformer; the envelope 104 of the output 102 approximates a low-frequency signal, such as one produced by a magnetic transformer. FIG. 2 is a graph 200 of another type of output 202 produced by an electronic transformer. In this example, the output 202 does not maintain consistent polarity relative to a virtual ground 204 within a half 60 Hz period 206. Thus, magnetic and electronic transformers behave differently, and a circuit designed to work with one may not work with the other.
For example, while magnetic transformers produce a regular AC waveform for any level of load, electronic transformers have a minimum load requirement under which a portion of their pulse-train output is either intermittent or entirely cut off. The graph 300 shown in FIG. 3 illustrates the output of an electronic transformer for a light load 302 and for no load 304. In each case, portions 306 of the outputs are clipped—these portions 306 are herein referred to as under-load dead time (“ULDT”). LED modules may draw less power than permitted by transformers designed for halogen bulbs and, without further modification, may cause the transformer to operate in the ULDT regions 306.
To avoid this problem, some LED light sources use a “bleeder” circuit that draws additional power from the halogen-light transformer so that it does not engage in the ULDT behavior. With a bleeder, any clipping can be assumed to be caused by the dimmer, not by the ULDT. Because the bleeder circuit does not produce light, however, it merely wastes power, and may not be compatible with a low-power application. Indeed, LED light sources are preferred over conventional lights in part for their smaller power requirement, and the use of a bleeder circuit runs contrary to this advantage. In addition, if the LED light source is also to be used with a magnetic transformer, the bleeder circuit is no longer necessary yet still consumes power.
Dimmer circuits are another area of incompatibility between magnetic and electronic transformers. Dimmer circuits typically operate by a method known as phase dimming, in which a portion of a dimmer-input waveform is cut off to produce a clipped version of the waveform. The graph 400 shown in FIG. 4 illustrates a result 402 of dimming an output of a magnetic transformer by cutting off a leading-edge point 404 and a result 406 dimming an output of an electronic transformer by cutting off a trailing-edge point 408. The duration (i.e., duty cycle) of the clipping corresponds to the level of dimming desired—more clipping produces a dimmer light. Accordingly, unlike the dimmer circuit for an incandescent light, where the clipped input waveform directly supplies power to the lamp (with the degree of clipping determining the amount of power supplied and, hence, the lamp's brightness), in an LED system the received input waveform may be used to power a regulated supply that, in turn, powers the LED. Thus, the input waveform may be analyzed to infer the dimmer setting and, based thereon, the output of the regulated LED power supply is adjusted to provide the intended dimming level.
One implementation of a magnetic-transformer dimmer circuit measures the amount of time the input waveform is at or near the zero crossing 410 and produces a control signal that is a proportional function of this time. The control signal, in turn, adjusts the power provided to the LED. Because the output of a magnetic transformer (such as the output 402) is at or near a zero crossing 410 only at the beginning or end of a half-cycle, this type of dimmer circuit produces the intended result. The output of electronic transformers (such as the output 406), however, approaches zero many times during the non-clipped portion of the waveform due to its high-frequency pulse-train behavior. Zero-crossing detection schemes, therefore, must filter out these short-duration zero crossings while still be sensitive enough to react to small changes in the duration of the intended dimming level.
Because electronic transformers typically employ a ULDT-prevention circuit (e.g., a bleeder circuit), however, a simple zero-crossing-based dimming-detection method is not workable. If a dimmer circuit clips parts of the input waveform, the LED module reacts by reducing the power to the LEDs. In response, the electronic transformer reacts to the lighter load by clipping even more of the AC waveform, and the LED module interprets that as a request for further dimming and reduces LED power even more. The ULDT of the transformer then clips even more, and this cycle repeats until the light turns off entirely.
The use of a dimmer with an electronic transformer may cause yet another problem due to the ULDT behavior of the transformer. In one situation, the dimmer is adjusted to reduce the brightness of the LED light. The constant-current driver, in response, decreases the current drawn by the LED light, thereby decreasing the load of the transformer. As the load decreases below a certain required minimum value, the transformer engages in the ULDT behavior, decreasing the power supplied to the LED source. In response, the LED driver decreases the brightness of the light again, causing the transformer's load to decrease further; that causes the transformer to decrease its power output even more. This cycle eventually results in completely turning off the LED light.
Furthermore, electronic transformers are designed to power a resistive load, such as a halogen bulb, in a manner roughly equivalent to a magnetic transformer. LED light sources, however, present smaller, nonlinear loads to an electronic transformer and may lead to very different behavior. The brightness of a halogen bulb is roughly proportional to its input power; the nonlinear nature of LEDs, however, means that their brightness may not be proportional to their input power. Generally, LED light sources require constant-current drivers to provide a linear response. When a dimmer designed for a halogen bulb is used with an electronic transformer to power an LED source, therefore, the response may not be the linear, gradual response expected, but rather a nonlinear and/or abrupt brightening or darkening.
In addition, existing analog methods for thermal management of an LED involve to either a linear response or the response characteristics of a thermistor. While an analog thermal-management circuit may be configured to never exceed manufacturing limits, the linear/thermistor response is not likely to produce an ideal response (e.g., the LED may not always be as bright as it could otherwise be). Furthermore, prior-art techniques for merging thermal and dimming level parameters perform summation or multiplication; a drawback of these approaches is that an end user could dim a hot lamp but, as the lamp cools in response to the dimming, the thermal limit of the lamp increases and the summation or multiplication of the dimming level and the thermal limit results in the light growing brighter than the desired level.
Therefore, there is a need for a power-efficient, supply-agnostic LED light source capable of replacing different types of existing bulbs, regardless of the type of transformer and/or dimmer used to power and/or control the existing bulb.